FINDING NONDESTRUCTIVE PARAMETERS FOR ROOT-TO-SHOOT RATIOS IN
DOUGLAS-FIR, GRAND FIR, AND REDWOOD SAPLINGS IN NORTHWEST
CALIFORNIA FOR BIOMASS AND CARBON STORAGE ESTIMATES
By
Walter Aron Kast
A Thesis Presented to
The Faculty of Humboldt State University
In Partial Fulfillment of the Requirements for the Degree
Master of Science in Natural Resources: Forestry, Watershed & Wildland Sciences
Committee Membership
Dr. John-Pascal Berrill, Committee Chair
Dr. Abeer Hasan, Committee Member
Dr. Lucy Kerhoulas, Committee Member
Dr. Alison O’Dowd, Graduate Coordinator
May 2017
ABSTRACT
FINDING NONDESTRUCTIVE PARAMETERS FOR ROOT-TO-SHOOT RATIOS IN
DOUGLAS-FIR, GRAND FIR, AND REDWOOD SAPLINGS IN NORTHWEST
CALIFORNIA FOR BIOMASS AND CARBON STORAGE ESTIMATES
Walter A. Kast
There is a need for better understanding of how woody biomass is allocated above
and belowground and how this allocation might differ among tree species. In this field of
research, investigators face challenges such as the laborious task of removing trees from
the soil with destructive sampling, and the cleaning, drying, and weighing of
belowground biomass (BGB). Therefore, researchers and practitioners most often rely on
existing models to predict BGB from easily-measurable aboveground variables such as
stem diameter and height. Such models have been developed for many tree species, but
commonly these models require inputs of diameter at breast height (dbh) and are not
designed to make predictions for younger saplings (i.e., below 5 cm dbh). To fill
knowledge gaps in young conifer BGB allocation, we studied three conifers native to the
north coast of California: coast redwood (Sequoia sempervirens), coast Douglas-fir
(Pseudotsuga menziesii var menziesii), and grand fir (Abies grandis). We sought to
determine: (i) Does the root-to-shoot ratio differ between the three species Douglas-fir,
grand fir, and coast redwood in afforestation plots? (ii) Does the root-to-shoot ratio of the
three species differ according to age (i.e. sampling across a span of three years?) (iii)
Does the competing flora alter the root-to-shoot ratio of any of the three species? (iv)
ii
What are the best “easily-measurable” aboveground variables to be included in prediction
equations for BGB in the three tree species?
Experimental plots were planted in 2008/09, and another in 2009/10 at the L. W.
Schatz Demonstration Tree Farm located in Humboldt County, CA. Five species were
planted: coast redwood, coast Douglas-fir, grand fir, red alder (Alnus rubra), and black
cottonwood (Populus balsamifera). Redwood, Douglas-fir, and grand fir were
destructively sampled for BGB measurement. A random sample of these three species
were excavated by hand, and separated into three sections: stems, roots and branches.
Each species had 24 trees sampled across the 3 years of data collection for a total of 72
trees. The sapling biomass components were weighed, dried in an oven, and re-weighed
to determine bone dry weight and root-to-shoot biomass ratios.
Before final root-to-shoot ratios and BGB models were created, auxiliary models
were developed to predict the weight of any roots that were broken off during the
excavation of the saplings. Models for severed root weight were tested against sapling
height, average crown width, lower crown base height, and stem diameter. Results
showed high correlation between root weight and stem diameter at ground line (caliper,
mm). Exponential models made the best predictions of weight of individual pieces of
broken root for all three species: Douglas-fir (R2= 0.86), grand fir (R2= 0.91), and
redwood (R2= 0.79).
After missing root weights had been predicted for each broken root on the root
system of each sample tree, summed, and added to the overall root mass, equations to
iii
predict BGB were developed and tested. Multivariable models were tested for all three
species, but showed no statistical significance. Bivariate regressions of BGB as function
of tree height (cm), average crown width (cm), lower crown base height (cm), stem
diameter (mm), year, and percent cover of competing flora were tested. In speciesspecific bivariate regressions, tree height, average crown width, and stem diameter were
all found to be statistically significant predictors of BGB for all three species. Douglas-fir
BGB was best predicted with a linear model utilizing caliper as the explanatory variable
(R2= 0.77). Grand fir BGB was also predicted well by a linear model with caliper as the
explanatory variable (R2= 0.92). Redwood BGB exhibited an exponential relationship
with caliper (R2= 0.91).
Root-to-shoot ratios for the three species averaged between 0.27 and 0.46. All
variables tested for BGB were also tested as predictor of root-to-shoot ratios, however for
Douglas-fir and grand fir, no significant relationships between root-to-shoot ratio and the
candidate predictor variables were found. For redwood, stem diameter, average crown
width, and sapling height all were significant predictors of root-to-shoot ratio. Redwood
sapling height was the best predictor of root-to-shoot ratio (R2= 0.37). For all three
species, ANOVA tested for differences in root-to-shoot ratios among sample ages. The
youngest Douglas-fir saplings (three years old) had higher root-to-shoot ratio than the
five and seven year old trees. Grand fir showed no differences in root-to-shoot ratios
according to age. Redwood root-to-shoot ratios were significantly different between ages
three and four, between ages four and five, and between ages four and six years old.
iv
ACKNOWLEDGEMENTS
The research that has been done in this paper has been funded by the L.W. Schatz
Demonstration Tree Farm. There was a large list of people that assisted in the research
and helped me reach my goal. My graduate advisor Dr. John-Pascal Berrill played a key
role in assisting me with research, and helped me problem solve various situations. His
guidance and professionalism are traits I will keep with me throughout my working
career, and personal life. Thank you for helping me realize and demonstrate my abilities
and potential. My committee members Dr. Abeer Hasan and Dr. Lucy Kerhoulas
provided key support with laboratory and academic resources and guidance. I also want
to personally thank George Pease, and Dr. Jeffry Kane for allowing me utilize the
equipment that allowed this research to happen. I also thank Gordon Schatz for guidance
and company when collecting data at the tree farm. Other people who helped in the field
and laboratory include but are not necessarily limited to: Jeffrey Ortiz, Keath Sakihara,
Jeffrey Paulson, Andrew Dobbs, Heesung Woo, Gabriel Goff, Nicholas Kilgore, Greg
Winkley, Christopher Crowell, Joel Bisson, Andrew Slack, and Anil Kizhakkepurakkal.
Finally, I want to thank my parents Rose and Georg Kast for their persistence, love,
patience, and overwhelming support of my goals even through rough times.
v
TABLE OF CONTENTS
ABSTRACT ........................................................................................................................ ii
ACKNOWLEDGEMENTS ................................................................................................ v
TABLE OF CONTENTS ................................................................................................... vi
LIST OF FIGURES ......................................................................................................... viii
LIST OF TABLES ............................................................................................................. ix
1. INTRODUCTION .......................................................................................................... 1
2. METHODS ..................................................................................................................... 6
2.1 Site Description ......................................................................................................... 6
2.2 Sapling Excavation and Data Collection ................................................................... 8
2.3 Missing Root Measurements ................................................................................... 11
2.4 Analysis of Factors Influencing Belowground Biomass and Root-to-Shoot Ratio 11
3. RESULTS ..................................................................................................................... 13
3.1 Missing Root Models .............................................................................................. 13
3.2 Tree Components .................................................................................................... 13
3.3 Belowground Biomass (BGB) Models ................................................................... 17
3.4 Root-to-Shoot Ratio Models ................................................................................... 21
4. DISCUSSION ............................................................................................................... 26
vi
4.1 Belowground Biomass (BGB) ................................................................................ 26
4.2 Root-to-shoot Ratios ............................................................................................... 27
4.3 Conclusions ............................................................................................................. 30
5. REFERENCES ............................................................................................................. 32
vii
LIST OF FIGURES
Figure 1.) Excavation techniques and data collection of 3 to 7-year-old saplings planted at
the L.W. Schatz Demonstration Tree Farm, Humboldt County, CA. A.) Visual
description of competing flora in vicinity of each sapling on study site. B.) Manual
excavation technique capturing as many coarse roots as possible. C.) Example of
excavated sapling (grand fir), before being separated into stem, branch, and root
components. D.) Measuring small end diameter of roots >2 mm that broke during
excavation. ................................................................................................................. 10
Figure 2.) Exponential relationship between broken root weight (g), and measured largeend diameter (mm) for Douglas-fir, grand fir, and redwood collected at the L.W.
Schatz Demonstration Tree Farm, Humboldt County, CA. Note: different axis scales
for each species depict different ranges of sample data. ........................................... 15
Figure 3.) Percent representation of average above- and belowground biomass
components from 3 to 7-year-old conifer saplings planted at L.W. Schatz
Demonstration Tree Farm, Humboldt County, California. ....................................... 16
Figure 4.) (A.) Mean root weight of Douglas-fir, grand fir, and redwood, with 95%
confidence intervals. (B.) Mean root-to-shoot ratio for Douglas-fir, grand fir, and
redwood with 95% confidence intervals. Data collected at L.W. Schatz
Demonstration Tree Farm, Humboldt County, CA. .................................................. 18
Figure 5.) Relationship between belowground biomass (root weight; g), and tree size
(caliper; mm) for age 3 to 7-year-old Douglas-fir, grand fir, and redwood saplings
planted at L.W. Schatz Demonstration Tree Farm, Humboldt County, CA. Note:
different axis scales for each species depict different ranges of sample data. .......... 20
Figure 6.) Age effect on root-to-shoot ratio; includes models for redwood, grand fir, and
Douglas-fir. Interaction variable tested showed no significant difference in slope
among species. ........................................................................................................... 24
viii
LIST OF TABLES
Table 1.) Candidate broken root weight models – relationship between root weight (W;
g), large-end diameter (D; mm), and model fit statistics for Douglas-fir (DF), grand
fir (GF) and redwood (RW) sapling roots terminating at 2 mm small-end diameter.
Best model shown in bold. ........................................................................................ 14
Table 2.) Candidate belowground biomass models – relationship between belowground
biomass (B; g) and stem caliper (C; mm) or stem height (H; cm) or average crown
width (CW; cm) or sapling age (A; years) and model fit statistics for Douglas-fir
(DF), grand fir (GF) and redwood (RW) sapling roots terminating at 2 mm small-end
diameter, including biomass predicted for missing roots. ......................................... 23
Table 3.) Candidate root-to-shoot models – relationship between root-to-shoot biomass
ratio (R:S) and stem caliper (D; mm) or stem height (H; cm) or average crown width
(CW; cm) or sapling age (A, years) and model fit statistics for Douglas-fir (DF),
grand fir (GF) and redwood (RW), and interaction between age and species (A&S1,
A&S2)........................................................................................................................ 25
Table 4.) Range of root-to-shoot ratios (R:S) for individual tree species, and average R:S
for common forest types found in the United States and Canada. Adapted from
Namm (2012)............................................................................................................. 28
ix
1
1. INTRODUCTION
Trees sequester CO2 from the atmosphere and store it as carbon in above and
belowground biomass. Since CO2 is a greenhouse gas, trees have the potential to offset
CO2 emissions causing climate change. We need better understanding of relationships
between carbon, tree size, and forest growth to inform forest management and policy
makers.
With growing concerns about greenhouse gas emissions and ways to mitigate for
the harmful effects, we need to find solutions for rising atmospheric carbon that can be
implemented quickly. In response to this need, in 1992 several countries agreed to the
United Nations Framework Convention on Climate Change (UNFCCC). The UNFCCC
had the goal of developing inventories of greenhouse gas emissions and sinks, or places
carbon could be stored. They also wanted to find ways to increase sinks and lower the
overall net greenhouse gas emissions (FAO, 2001). In 1997, the UNFCCC agreed on the
Kyoto Protocol to reduce greenhouse gas emissions to < 5% 1990 levels by 2012. This
protocol allowed countries that emit more carbon than the agreed upon limit to purchase
carbon offsets from countries or areas that have carbon sinks.
There are three ways that forest managers can help increase carbon sinks and
produce carbon credits to sell. The first way carbon can be stored is carbon sequestration
which includes afforestation, reforestation, restoration of damaged lands, and improved
silvicultural techniques to increase tree growth rate. Carbon conservation is the second
approach which includes conservation of biomass and carbon in soils, improved
2
harvesting techniques to maximize wood processing efficiency, and better use of
harvesting residuals from typical burning practices. The last way carbon can be handled
is carbon substitution that is intended to convert forest biomass into durable wood
products that will replace other energy dependent products, increased use of waste
material for biofuels and bioenergy production, and carbon storage in live biomass
plantations before use as biofuel (Montagnini & Nair, 2004). The carbon conservation
has potential for fast mitigation; however there is a cap that will likely be reached in
conservation. For more long term mitigation carbon sequestration would be the more
favorable option. Afforestation and reforestation both promote carbon sequestering and
storage in long-lived organisms (Trabucco et al., 2008).
For carbon assessment in forests and the modeling and sale of carbon credits, we
must accurately estimate the biomass in each tree. There are numerous equations and
methods to estimate aboveground biomass (AGB) for hardwoods and conifers (Cairns et
al., 1997). Unfortunately, there has been less research into estimating belowground
biomass (BGB) in root systems of trees due to the difficulty of sampling. Estimating root
biomass from aboveground measurements to circumvent destructive sampling will help
economically and efficiently predict carbon storage in trees (Nielson & Hansen, 2006). A
better understanding of belowground carbon sequestration and storage should give policy
makers confidence to include tree roots in carbon calculations and the sale of carbon
offsets.
3
Root biomass allocation has been studied in several tree species. It is variable
among and within species. For example, Namm (2012) extracted 10 adult tanoaks
(Notholithocarpus densiflorus) root systems to determine root-to-shoot ratios of single or
multi-stemmed tanoak. He reported root-to-shoot biomass ratio between .11 and .65
showing the wide range of ratios that can occur in one species. Tanoak root-to-shoot ratio
varied according to tree size and stand density. Root morphology models were developed
to predict biomass of tanoak roots lost during excavation (Namm & Berrill, 2016). Van
Hees & Clerkx (2003) found 37-46% of biomass for pedunculate oak (Quercus robur)
and as high as 41-50% for beech (Fagus sylvatica) was found belowground.
Root biomass allocation patterns vary according to tree size and age. Monk
(1966) examined root systems for adult loblolly pine (Pinus taeda) to determine dry root
mass, and discovered that root-to-shoot ratios were lower in trees with larger DBH.
Jenkins et al. (2003) showed that root biomass allocation decreases sharply as tree size
increases to 10 cm DBH, then decreases slowly as tree size increases further. Unknown is
how competing flora may help or hinder a species belowground carbon storage ability
(Monk, 1966). More research into root biomass allocation in younger trees is needed to
better understand how carbon storage might change throughout the life of the tree
(Brown, 2002). This information will allow for better understanding of how much and
where carbon is stored in young forests. Accurate total biomass measurements will
inform studies of carbon dynamics in reforestation and afforestation projects. The
difficulty and cost of destructive sampling for root biomass motivates us to develop
4
regression equations that predict BGB and root-to-shoot ratios in younger trees at
different ages.
According to the California State Air Resources Board (ARB), two methods are
in use to account for BGB in forest carbon offset programs. All states besides California,
Oregon, and Washington use the United States Forest Service’s Component Ratio
Method (CRM) (California Air Resource Board, 2013a). The CRM estimates volume in
the non-bole components of the tree as ratios of total AGB. These volumes are summed
to give total AGB and BGB converted to tons of carbon dioxide equivalents (CO2e) on a
per acre basis. For California, Oregon, and Washington, various equations are used to
predict bole volume and total AGB (California Air Resource Board, 2013b). The
estimates are then converted to CO2e and summed for all trees to give AGB on a per acre
basis. For BGB in these states, the Cairns equation’s (Cairns et al., 1997) must be used to
calculate CO2e. Cairns et al. (1997) used an assemblage of data from the literature to
develop regression equations to predict BGB. A problem with this method that Cairns et
al. (1997) discuss is that this approach can lead to an overestimation of BGB as high as
20%. This lessens our confidence in predictions of BGB and suggests more destructive
sampling and model building is needed.
This thesis describes a study of BGB in three commercially and ecologically
important conifers native to north coastal California: coast redwood (Sequoia
sempervirens), coast Douglas-fir (Pseudotsuga menziesii var. menziesii), and grand fir
(Abies grandis). The objective was to study factors hypothesized to affect BGB and root-
5
to-shoot ratio, then develop predictive models allowing users to estimate BGB from
easily-measurable aboveground variables. Variables hypothesized to affect BGB and
root-to-shoot ratio were tree species, size, age, and weed competition in young trees
excavated from two mixed plantations in different years.
6
2. METHODS
2.1 Site Description
The conifer saplings were excavated from two mixed plantations located at the
L.W. Schatz Demonstration Tree Farm (LWSDTF) 40°46'06.7"N 123°52'12.6"W,
approximately 40 km inland of Arcata in Humboldt County, California. The LWSDTF
covers approximately 150 hectares and is owned and used by Humboldt State University
as an experimental forest and teaching aid for students and faculty.
Coastal fog rarely reaches inland as far as the LWSDTF, so the Mediterranean
climate of the LWSDTF has warmer dry summers reaching average highs of 25.8 °C in
August. The average high and low temps for the coolest month are 0.3 and 12.2 °C for
January. Annual rainfall in the region averages 1010 mm and typically falls in the winter
months of November through March (Western Regional Climate Center, 2015).
The study site occurs within the Coastal Range Geomorphic Province and is
comprised mostly of the Franciscan Complex. The Franciscan Complex contains
“graywacke, shale, minor conglomerate, radiolarian chert and siliceous shale, minor
limestone, volcanic rocks, mafic-ultramafic putonic rocks, and their zeolite-to-blueschistfacies metamorphic equivalents” (Berkland et al., 1972).
The old-growth Douglas-fir forests were cleared from the area in the 1950s and
1960s, and grazing was attempted. Unstable geology and other factors such as unwanted
tree regeneration impacted pastoralism, prompting interest in restoring forest cover.
Tanoak and other hardwoods regenerated naturally. Sporadic natural regeneration of
Douglas-fir was supplemented by planting in an attempt to restore cover and productivity
7
using this merchantable native species. A few grand fir and hemlock had also regenerated
naturally and their advanced regeneration can now be found in the understory. Decades
after the original harvest, some redwoods were planted under the shade and shelter of
heavily thinned tanoak/fir stands and survive to this day. Unknown was whether redwood
plantations could be established in full sun, outside of their natural range for this location,
and how they would fare compared to the locally-adapted firs.
To compare survival and growth of planted redwood, Douglas-fir, grand fir, and
other species in mixed even-aged plantations, two experimental test plots were planted
within 2009; the first experimental test plot (site #1) was planted in February of 2009,
while the second experimental test plot (site #2) was planted one growing season later in
December of 2009. The first experimental test plot covered an area approximately 10 ×
40 m (33 × 132 ft). It was planted with five species: Douglas-fir, grand fir, coast
redwood, along with red alder (Alnus rubra), and black cottonwood (Populus
trichocarpa) that were not sampled for BGB. Seedlings were acquired though Smith
River Nursery, Hastings LLC. Ten replicates of each species (40 trees per species) were
planted in four-tree row plots with 1 m spacing between trees, and 2 m spacing between
each row resulting in 5,000 stems/ha (~2,000 tpa). The second experimental test plot also
covered an area approximately 10 × 40 m. The same replication, spacing, guard row, and
nursery stock was utilized; however, because of establishment failures in the first
experimental plot, bigleaf maple (Acer macrophyllum) was planted instead of cottonwood
(but it also fared poorly). Both sites were mowed prior to planting and in between data
collections.
8
2.2 Sapling Excavation and Data Collection
Data collection was done in the summer of 2013 when seedlings were just over three
years old ( i.e. >3 years since planting on site #2) or just over four years old (site #1),
then in the summer of 2014 when they were just over four or five years since planting. A
final data collection was performed when the trees were just over six years (site #2) or
seven years since planting (site #1), between March 2016 and August 2016. In each data
collection period, three random replicates of Douglas-fir, grand fir, and coast redwood
were collected in site #1, and six replicates of the same species in site #2. Measurements
of percent cover were made for vegetation type in the immediate vicinity; these included
Scotch broom (Cytisus scoparius), poison oak (Toxicodendron diversilobum), salal
(Gaultheria shallon), coyote brush (Baccharis pilularis), grasses, forbs, and self-seeded
conifers or hardwoods (Figure 1A). Sapling height (cm), live crown base height (cm),
stem diameter at ground line (caliper; mm), crown width (cm), and presence/absence of
browsing was recorded before excavation began. Each sapling was marked at the ground
line for reference after the sapling was removed from the soil. Shovels were used to
excavate as many of the roots as possible (Figure 1B). Maximum coarse root depth (mm)
was recorded after the sapling was removed. The excavated sapling was separated into
stem, branch, and root biomass components (Figure 1C). For each sapling, the stem was
removed from the roots at the mark of the ground line, and the branches removed from
the stem. All stems, branches and roots were placed into separate labeled weighed bags
and re-weighed to record the green weight. The bags were then placed in an oven set to
9
600C and weighed each day until the weight became constant, giving dry weights for
roots, stem, and crown biomass components.
10
Figure 1.) Excavation techniques and data collection of 3 to 7-year-old saplings planted at
the L.W. Schatz Demonstration Tree Farm, Humboldt County, CA. A.) Visual
description of competing flora in vicinity of each sapling on study site. B.) Manual
excavation technique capturing as many coarse roots as possible. C.) Example of
excavated sapling (grand fir), before being separated into stem, branch, and root
components. D.) Measuring small end diameter of roots >2 mm that broke during
excavation.
11
2.3 Missing Root Measurements
Coarse roots were defined as roots with diameter >2 mm. It was common to find
that some coarse roots broke during excavation. Once the root ball was separated from
the stem, terminal diameter of any broken roots >2 mm was recorded (Figure 1D). Pieces
of broken root that were recovered during excavation were stored separately for each
sapling. The entire root system for each sapling was weighed green, then dried and
weighed repeatedly until it reached constant weight. Individual roots were removed that
were complete and terminated at a diameter of 2 mm or smaller. The total dry weight of
each sapling’s entire root system (including the root ball and coarse roots) was recorded.
All available pieces of coarse roots for each species were placed back in the oven
(600C) until a stable weight occurred. Each root’s large-end diameter was measured to a
hundredth of a millimeter, then it’s weight was recorded. These data were used to
develop a regression model predicting the mass of any piece of root based on its largeend diameter. Linear and exponential models were created to account for the missing root
mass, beyond the end of severed root ends >2 mm diameter. These models allowed for a
better representation of total BGB of the saplings by replacing coarse root biomass lost
during the excavation process with a prediction of biomass lost for each root that broke at
a diameter > 2 mm.
2.4 Analysis of Factors Influencing Belowground Biomass and Root-to-Shoot Ratio
The below ground biomass (g) and root-to-shoot biomass ratio for each sapling
was regressed against the following candidate explanatory variables: percent cover of
12
competing flora (grass or shrub cover (%)), tree size (caliper (mm)), height (cm), live
crown base height (cm), crown width (cm), tree age, and a categorical species variable to
test for significant differences among Douglas-fir, grand fir, and redwood. Data were
transformed to reduce skewness in distributions. Linear, polynomial, exponential, and
power models were fitted using R 3.2.2 (R Foundation for Statistical Computing, 2015).
The models were compared in terms of goodness of fit by taking the square root of the
absolute residuals to determine error, normality was assessed through Q-Q plots along
with Anderson-Darling normality tests, and influential outliers were identified by plotting
Cook’s distance. After testing sapling age through a linear model as a continuous variable
assessing if age was significant, ANOVA and a TukeyHSD test were used to test for
differences between sample ages. To determine the best predicters of BGB and root-toshoot ratio for each species, competing models were compared in terms of Akaike
information criterion with correction for small sample size (AICc). Model comparison
information was reported in terms of Bayesian information criterion (BIC), R2, Adjusted
R2, standard error of residuals, and the number of influential outliers. An influential
outlier was defined as an observation with a Cook’s distance > 1.
13
3. RESULTS
3.1 Missing Root Models
Regressing the weight and large-end diameter measured on individual pieces of
coarse root gave models designed to predict root mass lost during excavation. A total of
75 Douglas-fir roots, 68 grand fir roots, and 55 redwood roots were measured. The
number of separate pieces differed by species because all available roots were measured
and weighed, and the Douglas-fir and grand fir saplings had more roots. Pieces of coarse
roots lost during excavation amounted to 1-22% of the total BGB. Exponential models fit
the data for root weight versus large-end diameter (LED) best and were selected for all
three species (Table 1, Figure 2). These exponential models can be used to predict weight
of a missing root in grams as a function of root diameter (LED; mm) at the broken end,
such that:
Douglas-fir missing root weight = 0.36e0.31LED.
(Eq.1)
Grand fir missing root weight = 0.92e0.21LED
(Eq.2)
Redwood missing root weight = 0.58e0.27LED
(Eq.3)
Missing root biomass accounted for 0.3 to 8 % of total BGB in redwood, 0.04 – 4.8% in
Douglas-fir and 4.5-9.8 % in grand fir.
3.2 Tree Components
On average, BGB represented up to 29% of the total tree biomass for the three
conifer species. Results also showed that allocation of biomass to roots was comparable
to stem biomass allocation in each of the species (Figure 3). Roots of individual Douglas-
14
Table 1.) Candidate broken root weight models – relationship between root weight (W;
g), large-end diameter (D; mm), and model fit statistics for Douglas-fir (DF), grand fir
(GF) and redwood (RW) sapling roots terminating at 2 mm small-end diameter. Best
model shown in bold.
Species
DF1
Equation
W= 0.36e
0.31D
DF4
W=-1.67+1.35D0.22D2+0.02D3
W=1.310.471D+0.106D2
W=-1.86+0.802D
GF1
W=0.92e0.21D
GF2
W =-1.79 +1.61D0.21D2+0.01D3
W=4.131.89D+0.16D2
DF2
DF3
GF3
AIC
166.73
BIC
173.60
R2
0.87
Adj. R2
0.86
Residual SE
0.74
Outliers
0
170.50
181.95
0.87
0.86
0.75
5
174.98
184.14
0.85
0.85
0.78
7
200.65
207.52
0.79
0.78
0.93
4
248.54
255.16
0.91
0.91
1.50
0
249.70
260.73
0.92
0.91
1.50
4
262.25
271.10
0.90
0.89
1.65
5
GF4
W=-4.1+1.44D
329.48
336.10
0.71
0.70
2.75
4
RW1
W= 0.58e0.27D
143.18
148.97
0.80
0.79
0.95
0
RW2
W=1.190.33D+0.10D2
W=2.46-1.13D+
0.25D2-0.01D3
141.77
149.50
0.81
0.80
0.93
3
143.24
152.90
0.81
0.80
0.93
2
156.43
162.23
0.74
0.73
1.08
3
RW3
RW4
W= -1.63+0.85D
15
Figure 2.) Exponential relationship between broken root weight (g), and measured largeend diameter (mm) for Douglas-fir, grand fir, and redwood collected at the L.W. Schatz
Demonstration Tree Farm, Humboldt County, CA. Note: different axis scales for each
species depict different ranges of sample data.
16
Figure 3.) Percent representation of average above- and belowground biomass
components from 3 to 7-year-old conifer saplings planted at L.W. Schatz Demonstration
Tree Farm, Humboldt County, California.
17
fir saplings comprised 14-32% of whole tree biomass. Grand fir roots represented a high
of 37% and a low of 17% of BGB, respectively. Redwood roots comprised the highest
proportion of total tree weight, with BGB ranging from 26-44% of totals. Branches and
foliage were the heaviest of the three tree biomass components for all three species.
Douglas-fir branches and foliage represented a high of 69% and a low of 36% of total
tree weight. Grand fir branches and foliage represented between 39% and 63% of total
tree biomass. The weight of redwood branches and foliage was least variable, ranging
from 37% to 52% of total tree biomass in saplings aged 3 to 7 years old.
3.3 Belowground Biomass (BGB) Models
A total of 24 saplings of each species were sampled for BGB of coarse roots
(Figure 4). Generalized linear models with ‘Species’ as a categorical variable were tested
for all three species, and showed that all species had significantly different BGB. A
comparison of model AICc among models with one or more explanatory variables
indicated that bivariate models were best. For all three species, bivariate models indicated
that BGB was positively correlated with stem diameter (mm), sapling height (cm), age
(years), or average crown width (cm) (P>0.0001). Results from the TukeyHSD test
showed BGB differed significantly only between years four and six for Douglas-fir and
redwood. Grand fir BGB was significantly different across all years, except year four was
not significantly different from any other year.
Douglas-fir BGB had strong correlation with sapling height and average crown
width. Each model had very similar AIC values that only differed by 0.22, as well as
residual standard error that differed by only 0.45 (Table 2). Each had almost the
18
Figure 4.) (A.) Mean root weight of Douglas-fir, grand fir, and redwood, with 95%
confidence intervals. (B.) Mean root-to-shoot ratio for Douglas-fir, grand fir, and
redwood with 95% confidence intervals. Data collected at L.W. Schatz Demonstration
Tree Farm, Humboldt County, CA.
19
same R2 (0.49 for average crown width and 0.48 for sapling height) showing that less
than 50% of the variation in root mass was explained by sapling height or average crown
width. The best predictor of BGB was stem diameter (caliper). Two separate models, one
linear model and one exponential model, used caliper as a predictor variable (Table 2).
Again, the difference between these models was minimal with both AICs for the linear
and exponential model equaling 270.78. The residual standard errors were within 0.01,
but when fitting and evaluating the linear model, an outlier was detected. The exponential
model did not result in identification of influential outliers and appeared to fit the data
well (Figure 5), so it was determined to be the best model.
Grand fir BGB also had high correlations with stem diameter at ground line
(caliper), sapling height, and average crown width. Models of grand fir BGB as a
function of either height or average crown width had similar R2, with adjusted R2 that
were within 0.1 of each other, and similar residual standard errors (Table 2). Unlike
Douglas-fir BGB, a linear model of grand fir sapling caliper worked better than the
exponential model (Table 2, Figure 5), and was identified as the best BGB model.
Belowground biomass for redwood was best predicted by sapling stem diameter
(caliper). An exponential model fit the redwood BGB-caliper data best (Table 2, Figure
5).
20
Figure 5.) Relationship between belowground biomass (root weight; g), and tree size
(caliper; mm) for age 3 to 7-year-old Douglas-fir, grand fir, and redwood saplings planted
at L.W. Schatz Demonstration Tree Farm, Humboldt County, CA. Note: different axis
scales for each species depict different ranges of sample data.
21
3.4 Root-to-Shoot Ratio Models
A GLM of root-to-shoot ratio as a function of species (categorical variable) and
age (continuous variable) indicated that the ratio differed significantly among species
(Figure 4), and according to age. The highest root-to-shoot ratios were recorded for the
youngest (age 3) redwood, Douglas-fir, and grand fir (Table 3). Using the same model,
interactions were tested revealing that there were no significant differences in slope
between the three species (Figure 6). Counter to our expectations, competing flora did not
have a detectable influence on the root-to-shoot ratio or BGB of conifer saplings planted
at LWSDTF.
The average root-to-shoot ratio for all 24 Douglas-fir saplings was 0.27, with a
high of 0.47, and a low of 0.16. Ratios for Douglas-fir tended to fall with an increase in
aboveground biomass. Root-to-shoot ratios were tested against age with an ANOVA to
determine if there was a difference between them. When the saplings were four years old,
the root-to-shoot ratio was significantly higher than in the five and seven-year-old trees.
There were no other statistically significant differences between the other sample ages.
All variables used in BGB equations were tested as predictors of root-to-shoot ratios,
however only the model with age was statistically significant (P<0.05) (Table 3).
Grand fir had a 10% higher average root-to-shoot ratio than Douglas-fir at 0.37,
with a high of 0.58, and a low of 0.20 for all 24 trees. However, after testing with
ANOVA for sapling age and root-to-shoot ratios, there were no statistically significant
22
differences found. Again, just like Douglas-fir, age was the only variable that was
significant in predicting root-to-shoot ratios (Table 3).
Redwood had the highest root-to-shoot ratios for all 24 trees with an average of
.46, a high of .78, and a low of just .36. Redwood root-to-shoot ratio was also most
variable across all sample ages. Using ANOVA, redwood root-to-shoot ratios were
significantly different between ages 4-5, 5-6, and 5-7, but not 3-4 or 6-7 years old.
Unlike the other two species however, redwood stem diameter, sapling height, average
crown width, and age were all statistically significant predictors of root-to-shoot ratio.
Sapling height was the best predictor of root-to-shoot ratio in redwood saplings,
indicating that root-to-shoot ratios was lower among taller redwood saplings (Table 3).
23
Table 2.) Candidate belowground biomass models – relationship between belowground
biomass (B; g) and stem caliper (C; mm) or stem height (H; cm) or average crown width
(CW; cm) or sapling age (A; years) and model fit statistics for Douglas-fir (DF), grand fir
(GF) and redwood (RW) sapling roots terminating at 2 mm small-end diameter, including
biomass predicted for missing roots.
AICc
BIC
R2
Adj. R2
Residual SE
Outliers
271.98
274.31
0.78
0.77
62.85
0
271.98
274.31
0.78
0.76
62.84
1
B=33.782+2.469CW
292.54
294.88
0.49
0.47
96.46
2
DF.H
B= -0.898+1.398H
292.76
295.10
0.48
0.46
96.91
1
DF.A
B=2.37+52.24A
302.23
304.56
0.24
0.20
118.00
2
GF.C1
B=-137.49 +15.682C
256.35
259.88
0.93
0.92
46.53
1
GF.C2
B=83.814e0.042C
273.78
277.31
0.84
0.83
66.91
1
GF.CW
B=-72.547+3.939CW
275.56
278.97
0.73
0.72
88.82
1
GF.H
B=-292.032+3.245H
285.06
288.59
0.75
0.74
84.63
2
GF.A
B=-245.35+106.91A
296.45
298.78
0.62
0.60
104.60
1
RW.C2
B= 35.764e0.0718C
254.04
257.58
0.92
0.91
44.35
2
RW.C1
B= -156.500+16.535C
262.40
265.93
0.88
0.88
52.78
1
RW.H
B= -275.564+2.932H
270.90
274.44
0.83
0.82
63.01
1
B = -58.036+3.215CW
273.02
276.55
0.82
0.81
65.85
2
B=-129.38+69.56A
305.46
307.80
0.32
0.29
126.30
1
Model
Equation
DF.C1
B=-152.310+12.935C
DF.C2
DF.CW
RW.CW
RW.A
B=11.369e
0.0058C
24
Figure 6.) Age effect on root-to-shoot ratio; includes models for redwood, grand fir, and
Douglas-fir. Interaction variable tested showed no significant difference in slope among
species.
25
Table 3.) Candidate root-to-shoot models – relationship between root-to-shoot biomass
ratio (R:S) and stem caliper (D; mm) or stem height (H; cm) or average crown width
(CW; cm) or sapling age (A, years) and model fit statistics for Douglas-fir (DF), grand fir
(GF) and redwood (RW), and interaction between age and species (A&S1, A&S2).
Model
Equation
AICc
BIC
R2
Adj. R2
DF.R:S.A
R:S=0.376+-0.014A
-75.51
-73.81
Residual SE
0.13
0.09
0.05
1
GF.R:S.A
R:S=0.314-0.010A
-68.74
-66.40
0.93
0.05
0.01
1
RW.R:S.H
R:S=0.417-0.000H
-85.37
-83.03
0.42
0.39
0.04
1
RW.R:S.C
R:S=0.386-.004C
-84.68
-82.34
0.41
0.38
0.04
1
RW.R:S.CW
R:S=0.362-0.0007CW
-82.13
-79.80
0.34
0.31
0.04
1
RW.R:S.A
R:S=0.376+-0.014A
R:S=0.330.06DF+0.04RW0.01A
R:S=0.310.03DF+0.06RW0.01GFA-0.01DFA0.004RWA
-75.51
-73.81
0.13
0.09
0.05
1
-233.30
-222.80
-
-
0.05
3
-228.80
-214.60
-
0.05
3
A&S1
A&S2
-
Outliers
26
4. DISCUSSION
4.1 Belowground Biomass (BGB)
We discovered for Douglas-fir, grand fir, and redwood there were easily
measured aboveground components that correlated strongly with BGB. For all three
species, we hypothesized that stem diameter, tree height, average crown width, live
crown base height, and competing flora would explain variations in root biomass. Each
species had the same variables correlating with sapling BGB. Predictive models
performed best when utilizing the stem diameter at ground line (caliper) of each sapling
to predict BGB. Out of the hypothesized variables only sapling age, stem diameter,
average crown width, and height were statistically significant predictor variables. At our
site in northern California, height was highly correlated with crown size (Pearson’s
correlation coefficient = 0.89, P-value < 0.0005) explaining why it also predicted BGB
well. This may not hold over a wider range of tree species. Vogt et al. (1983) found that
root biomass tended to increase with an increase in tree size until crown closure,
suggesting that this was due to a tree’s tendency to extend their roots to reach the edge of
crown “drip line”.
We did not test for effects of stand level variables on BGB. Other variations in
BGB can be explained by stand level variables. The study site from which the saplings
were excavated was a warm south facing slope receiving full sun. We did not sample
saplings from different sites with different aspects, or saplings in the understory of multiaged stands. Danjon et al. (2005) showed how root morphology can be altered by stand
27
level variables such as wind, soil, water table, and site class. This suggests further
research is needed to determine if other planted or naturally regenerated even-aged and
multi-aged stands follow the same trends as found in this paper reporting results for
conifers planted in a mixed plantation.
The expectation was that conifer saplings would have greater BGB with
advancing age. For Douglas-fir and redwood, no significant differences were found
according to tree age. For grand fir, BGB increased significantly each year except for
between year four and year five. High variation in tree size within each age sampled
likely prevented detection of differences according to tree age.
4.2 Root-to-shoot Ratios
Root-to-shoot ratios can vary widely within and between species and forest types
(Table 4). George et al. (1997) reported root-to-shoot ratios as high as 1, equal
distribution of biomass above and belowground, in three-year-old potted Scots pine and
Douglas-fir saplings grown in a temperature controlled green house (Table 4). Some
species have a wide range of root-to-shoot ratios when a broad range of tree sizes are
studied. For example, tanoak sampled across one 150 ha property had root-to-shoot ratios
of 0.11 through 0.65 (Namm, 2012). Others species were shown to have small variations,
such as balsam fir, which had ratios between 0.32 and 0.40 (Lavigne & Krasowski,
2007). In my study at the LWSDTF, the root-to-shoot ratios for Douglas-fir, grand fir,
and redwood planted 3 to 7 years before excavation were highly variable (Figure 6).
Root-to-shoot ratios decreased with advancing tree age, showing how coarse root growth
28
Table 4.) Range of root-to-shoot ratios (R:S) for individual tree species, and average R:S
for common forest types found in the United States and Canada. Adapted from Namm
(2012).
Species/Forest Type
R:S
Tree ages
Author/Sample Country
Abies balsamea
0.32-0.40
10-80 years
Lavigne & Krasowski, 2007
Abies grandis
0.20-0.58
3-7 years
Research results
Acer saccharum
0.18-0.47
12-39 years
Whittaker et al. 1974
Acer spicatum
0.29-0.31
19-260 years
Whittaker et al. 1974
Betula lutea
0.15-0.37
19-260 years
Whittaker et al. 1974
Betula pendula
0.34-0.48
2 years
Van Hees & Clerkx, 2003
Fagus grandifolia
0.15-0.47
19-260 years
Whittaker et al. 1974
Fagus sylvatica
0.04-0.13
44-114 years
Bolte et al. 2004
Fagus sylvatica
0.66-1.00
3 years
Van Hees & Clerkx, 2003
Notholithocarpus densiflorus
0.11-0.65
N/A
Namm, 2012
Picea abies
0.15-0.30
44-114 years
Bolte et al. 2004
Picea abies
0.68-0.72
3 years
George et al. 1999
Picea rubens
0.41-0.54
19-260 years
Whittaker et al. 1974
Pinus sylvestris
0.89-0.99
3 years
George et al. 1999
Pinus taeda
0.20-0.83
10+ years
Monk, 1966
Pseudotsuga menziesii
0.79-1.00
3 years
George et al. 1999
Pseudotsuga menziesii
0.14-0.32
3-7 years
Research results
Quercus robur
0.62-0.84
2 years
Van Hees & Clerkx, 2003
Sequoia sempervirens
0.23-0.59
1-4 years
Phillips et al. 2013
Sequoia sempervirens
0.26-0.44
3-7 years
Research results
Douglas-fir
0.21*
36-200 years
United States
Mesic deciduous hardwood
0.25*
43 years
United States
Mixed evergreen
0.53*
32 years
United States
Red alder
0.20*
Mature forest
United States
Spruce
0.23*
84-212 years
Canada
Subalpine coniferous
0.27*
70-78 years
United States
* Stand R:S (stand-level belowground biomass over the aboveground biomass) of forest types calculated
from values summarized by Cairns et al. (1997) in their review of above- and belowground biomass from
various biomes.
29
is a higher priority for young trees that eventually transition to allocating more growth to
aboveground components. Grand fir had a high of 0.58 and a low of 0.20, among the 24
saplings sampled. Douglas-fir ranged from 0.16 to 0.47, which was lower than the range
of values reported by George et al. (1997) for young Douglas-fir. Redwood had the
highest root-to-shoot ratio recorded of 0.78, a low of 0.36, which was higher than the
ratios reported by Phillips et al. (2013) for redwood ages 1-4 years old.
Douglas-fir ratios found at the study site were just slightly above that found by
Cairns et al. (1997). The grand fir and redwood sampled at LWSDTF both had much
higher root-to-shoot ratios than the average presented by Cairns et al. (1997) for the
mixed evergreen forest type.
Some differences in root-to-shoot ratio between trees of the same species can be
attributed to tree age. Monk (1966) found that root-to-shoot ratios tended to decline with
advancing tree age. This was consistent with the data collected at the LWSDTF for all
three species (Figure 6). The highest root-to-shoot ratios were observed in the first year
of data collection when the saplings were three or four years old. Douglas-fir did not
show a statistically significant difference in root-to-shoot ratios between ages four and
seven, but regression analysis revealed a decrease in the root-to-shoot ratio with
advancing age for all three species. George et al. (1997) and Monk (1966) showed that
root-to-shoot ratios changed most rapidly in the first few years of the tree’s life. Root-toshoot ratios in grand fir had yet to be reported, but by comparing the data for the
LWSDTF to root-to-shoot from the same genus (Abies balsamea), averages found at the
30
LWSDTF fell within the same range of values (Table 4). Average root-to-shoot ratios
reported by Phillips et al. (2013) for redwood also were within the same range found at
the LWSDTF, however this does not prove that location does not affect root-to-shoot
ratio. Large sample sizes would be needed to detect differences due to the inherent
variability in root-to-shoot ratios at a single site such as the LWSDTF. More data
covering a wide range of ages are needed to reliably model the effect of tree age on rootto-shoot ratios. Due to the greater difficulty of sampling larger older tree root systems,
any new research on root-to-shoot ratios for any species will be valuable and the raw data
should be shared to allow for meta-analysis across broader ranges of ages and geographic
areas and development of more robust predictive models for BGB and coarse root carbon.
4.3 Conclusions
Results of this research show that roots can represent almost as much as biomass
as the visible portion of young trees. Douglas-fir, grand fir, and redwood data for the five
sapling ages studied at the LWSDTF showed that root-to-shoot ratios declined with
advancing tree age; as conifers get older, their above ground components amass biomass
more rapidly than their root system.
Destructive sampling by excavation was laborious, and pieces of coarse roots lost
during excavation amounted to 1-22% of the total BGB. Predictive equations created
from destructive sampling data revealed high correlations between BGB and
aboveground variables. Stem diameter for all three species explained the most variations
in belowground root mass. Using these equations, forest landowners and managers can
31
obtain predictions of the amount of carbon stored belowground in young conifer
plantations, and how it increases as trees grow.
32
5. REFERENCES
Berkland, J. O., Raymond, L. A., Kramer, J. C., Moores, E. M., & O'Day, M. (1972).
What is Franciscan?. AAPG Bulletin, 56(12), 2295-2302.
Bolte, A., Rahmann, T., Kuhr, M., Pogoda, P., Murach, D., & Gadow, K. V. (2004).
Relationships between tree dimension and coarse root biomass in mixed stands of
European beech (Fagus sylvatica L.) and Norway spruce (Picea abies [L.]
Karst.). Plant and Soil, 264(1), 1-11.
Brown, S. (2002). Measuring carbon in forests: current status and future challenges.
Environmental Pollution, 116(3), 363-372.
Cairns, M., Brown, S., Helmer, E. et al. (1997) Root biomass allocation in the worlds
upland forests. Oecologia, 111,1-11. doi:10.1007/s004420050201.
California Air Resource Board. (2013a). Compliance Offset Protocol U.S. Forest
Offset Projects – Instructions (45states). California EPA. Available online at
www.arb.ca.gov/cc/capandtrade/protocols/usforest/instructions-45states.htm;
last accessed Oct. 4, 2016.
California Air Resource Board. (2013b). Compliance Offset Protocol U.S. Forest
Offset Projects – Instructions (CA, OR, WA). California EPA. Available
online at www.arb.ca.gov/cc/capandtrade/protocols/usforest/instructions-ca-orwa.htm; last accessed Oct. 4, 2016.
Danjon, F., Fourcaud, T., & Bert, D. (2005). Root architecture and wind-firmness of
mature Pinus pinaster. New Phytologist, 168(2), 387-400.
FAO (2001). State of the world’s forests. Food and Agriculture Organization of the
United Nations. Rome, Italy, 181 pp.
George, E., Seith, B., Schaeffer, C., & Marschner, H. (1997). Responses of Picea, Pinus
and Pseudotsuga roots to heterogeneous nutrient distribution in soil. Tree
Physiology, 17(1), 39-45.
Jenkins, J. C., Chojnacky, D. C., Heath, L. S., & Birdsey, R. A. (2003). National-scale
biomass estimators for United States tree species. Forest Science, 49(1), 12-35.
Lavigne, M. B., & Krasowski, M. J. (2007). Estimating coarse root biomass of balsam
fir. Canadian Journal of Forest Research, 37(6), 991-998.
33
Monk, C. (1966). Ecological importance of root/shoot ratios. Bulletin of the Torrey
Botanical Club, 93(6),402-406.
Monk, C. D. (1966). Root-shoot dry weights in loblolly pine. Botanical Gazette, 246-248.
Montagnini, F., & Nair, P. K. R. (2004). Carbon sequestration: an underexploited
environmental benefit of agroforestry systems. Agroforestry Systems, 61(1-3),
281-295.
Namm, B. H. (2012). Root morphology and belowground carbon storage in tanoak
(Notholithocarpus densiflorus) (Master’s Thesis, Humboldt State University).
Namm, B. H., & Berrill, J. P. (2016). Tanoak (Notholithocarpus densiflorus) Coarse Root
Morphology: Prediction Models for Volume and Biomass of Individual Roots.
Open Journal of Forestry, 6(1), 1-13.
Nielsen, C. C. N., & Hansen, J. K. (2006). Root CSA-root biomass prediction models in
six tree species and improvement of models by inclusion of root architectural
parameters. Plant and Soil, 280(1-2), 339-356.
Phillips, C. J., Marden, M., Lambie, S., Watson, A., Ross, C., & Fraser, S. (2013).
Observations of below-ground characteristics of young redwood trees (Sequoia
sempervirens) from two sites in New Zealand–implications for erosion
control. Plant and Soil, 363(1-2), 33-48.
Trabucco, A., Zomer, R. J., Bossio, D. A., van Straaten, O., & Verchot, L. V. (2008).
Climate change mitigation through afforestation/reforestation: a global analysis of
hydrologic impacts with four case studies. Agriculture, Ecosystems &
Environment, 126(1), 81-97.
Van Hees, A. F. M., & Clerkx, A. P. P. M. (2003). Shading and root–shoot relations in
saplings of silver birch, pedunculate oak and beech. Forest Ecology and
Management, 176(1), 439-448.
Vogt, K. A., Moore, E. E., Vogt, D. J., Redlin, M. J., & Edmonds, R. L. (1983). Conifer
fine root and mycorrhizal root biomass within the forest floors of Douglas-fir
stands of different ages and site productivities. Canadian Journal of Forest
Research, 13(3), 429-437.
Western Regional Climate Center, (2015). Desert Research Institute, Reno, Nevada.
Available online at http://www.wrcc.dri.edu/; last accessed on November 9th
2015.
34
Whittaker, R. H., Bormann, F. H., Likens, G. E., & Siccama, T. G. (1974). The Hubbard
Brook ecosystem study: forest biomass and production. Ecological
Monographs, 44(2), 233-254.